Why Canada's Hubs for AI Research Are Attracting Global Tech

ARTIFICIAL INTELLIGENCE INCUBATORS Why Canada’s Hubs for AI Research Are Attracting Global Tech

OCTOBER 2018 Paul Briggs Contributors: Angela Kim, Victoria Petrock ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH

Canada is among the pioneers when it comes to AI research and development. It’s home to a bustling scene that pairs academia with industry, with the aid of government funding. Machine learning is touching a variety of disciplines, but the marketing field is emerging as the primary area for designing real-world business applications.

■■ Why is Canada attracting global tech companies Marketing Is Top Focus for Artiﬁcial Intelligence for artificial intelligence research? Canada’s AI Firms in Canada, 2017 ecosystem is thriving. Particularly in Montreal, ranked by number of ﬁrms Toronto and Edmonton, clusters of AI innovation have Marketing 38 developed to tie private enterprise to academia, where advanced machine-learning research attracts top talent. Fintech 20 Health/pharmaceuticals 18 ■■ What do consumers think of their engagement Human resources 14 with artificial intelligence? Consumer perceptions of their engagement with AI is low, likely due to the Social media 11 behind-the-scenes nature of the technology. Algorithms 10 Customer service for pervasive consumer services from Facebook, Lifestyle 9 Google, Amazon and others rely on AI to provide Media 9 more personalized experiences, but a large portion of 8 AI-as-a-Service consumers do not see the connection. 7 Cybersecurity

■■ Which industry uses artificial intelligence Source: Green Technology Asia Pte Ltd, "The Canadian AI Ecosystem: A effectively? Applications in marketing have become the 2018 Proﬁle," Feb 12, 2018 most mature business AI deployments. Programmatic 240347 www.eMarketer.com advertising channels deliver optimization built with KEY STAT: AI startups cover a range of industries, but machine learning, chatbots have become a proven marketing is the top area for machine-learning testing and method of customer engagement and recommendation development, with no less than 38 firms working on AI engines power better ecommerce experiences. for marketing solutions in Canada.

WHAT’S IN THIS REPORT? This report looks at the why and how of Canada’s thriving AI scene. It covers the research and development infrastructure, consumer attitudes toward machine learning and the top applications for brands and marketers. CONTENTS

2 Artificial Intelligence Incubators: Why Canada’s Hubs for AI Research Are Attracting Global Tech 3 A Northern AI Incubator 4 Consumer Perspectives on AI 6 Real-World AI for Marketers 12 Key Takeaways 12 eMarketer Interviews 13 Read Next 13 Sources 13 Editorial and Production Contributors

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 2 A NORTHERN AI INCUBATOR These announcements prompted Google’s executive chairman Eric Schmidt to tweet last fall that Canada “quadrupled down” on AI with a smart mix of four key Artificial intelligence (AI) is touching every industry, elements: government, universities, large companies as technologists begin shaping practical applications and startups. and executives attempt to understand how it will influence profitability and competitive differentiation. The City of Montreal has drawn significant investment More than just a business opportunity, consumers from global tech giants interested in cutting edge AI are also being educated about AI’s potential—and research and talent. These moves represent a broader strategy by corporate leaders to establish close ties with potential threats, like rampant automation resulting academia, especially in advanced computing disciplines in the loss of certain types of jobs. like AI. Consider the following announcements in the past couple of years to situate corporate research centers near The specter of AI isn’t new. After all, Steven Spielberg’s the cluster of innovation emanating from the Universities film “AI” was released in 2001. A couple of decades of Montreal and McGill: before then, in 1983, The Canadian Institute for Advanced Research (CIFAR) created a group called Artificial ■■ Facebook: opened a FAIR (Facebook Artificial Intelligence, Robotics and Society. It was led by Geoffrey Intelligence Research) lab in Montreal in Hinton, a University of Toronto-based AI pioneer (Hinton September 2017 now works for Google in addition to his duties at the university). ■■ Google: Google Brain opened a lab in November 2016, led by Montreal-based deep learning scientists Back then, AI had been heavily researched in academia, but computing power fell short of what was required ■■ Microsoft: acquired Maluuba in January 2017, a for meaningful machine learning (a synonym for AI). Montreal-based AI lab The advent of cloud computing, big data systems and ■■ Samsung: Its Samsung Electronics Advanced Institute increased processing power in recent years has made of Technology (SAIT) opened a lab in September 2017. practical applications of AI a reality today. Companies like Google, Facebook, Apple, IBM and Amazon have ■■ Thales SA: announced plans to open a lab in Montreal built AI into their algorithms and services to deliver more in October 2017 personalization every day. In addition to industry commitment to the AI scene Canada is on a shortlist of global AI leaders, punching in Canada, the political landscape in North America above its weight due to AI pioneers like Hinton, Yoshua reinforces the sector’s strength. In 2016, there was a Bengio of the University of Montreal and startup noticeable “brain drain” of AI talent from Canada to the Element.ai and the University of Alberta’s Rich Sutton. US industry prior to the election of Donald Trump. But a These academics have built some of the finest AI degree survey of the membership of MaRS, the Toronto-based programs in the world. innovation incubator, reported that 53% of companies saw more international applicants in 2017 compared with In 2017, the federal government in Ottawa announced 2016, primarily from the US. The Trump administration’s the Pan-Canadian Artificial Intelligence Strategy, a repeal of the International Entrepreneur Rule, which CA$125 million five-year commitment to fund research enables immigrants to grow a business for up to five led by groups centered around the pioneering scientists: years without visa expiry, has made Canada a more the Montreal Institute for Learning Algorithms (MILA), The attractive place for skilled technologists. Alberta Machine Intelligence Institute (AMII) in Edmonton, and the Vector Institute for Artificial Intelligence, based The buzz surrounding the AI sector in Canada appears in Toronto. to be influencing private industry’s openness to invest. Executives in Canada are increasingly bullish on AI Provincial governments are also joining the funding party. investment in the next three to five years—more so The Quebec government has provided $100 million than those in the US and the global average, according to its local AI community in Montreal and Ontario has to “AI Looks North: Bridging Canada’s Corporate committed $50 million for Vector. Artificial Intelligence Gap,” a May 2018 report from McKinsey & Company.

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 3 How Much Do Executives in Canada Expect Their AI Marketing Is Top Focus for Artiﬁcial Intelligence Investment to Change in the Next 3 Years? Firms in Canada, 2017 % of respondents, Dec 2017 ranked by number of ﬁrms

Signiﬁcant increase (10%-15%) 38% Marketing 38

Modest increase (0%-10%) 49% Fintech 20

Steady 10% Health/pharmaceuticals 18

3% Decrease Human resources 14

Note: n=120 Social media 11 Source: McKinsey & Company, "AI looks North: Bridging Canada's corporate artiﬁcial intelligence gap," May 14, 2018 10 Customer service 240268 www.eMarketer.com Lifestyle 9 Venture capital is readily available for AI startups in Media 9 Canada. PwC Canada reported that 2017 was a record 8 AI-as-a-Service year for AI investments in the country, with startups 7 Cybersecurity raising CA$191 million. PwC and CB Insights’ “Money Source: Green Technology Asia Pte Ltd, "The Canadian AI Ecosystem: A Tree” report showed that investments in Canadian AI 2018 Proﬁle," Feb 12, 2018 companies increased 88% year over year in Q1 2018, 240347 www.eMarketer.com with eight deals totaling CA$83 million. “The opportunity for AI experts is to simplify concepts In “The Pulse on Fintech 2018,” KPMG reported similar like machine learning and contextual computing for heavy venture capital funding and merger and acquisition non-computer scientists,” said Hossein Rahnama, CEO activity in Canada. “AI continued to be one of the and founder of Flybits, a Toronto-based AI fintech firm. dominant focus areas for investors in Canadian fintech. “This allows people to focus more on use cases, like Canada is increasingly regarded as a global leader in AI in marketing and customer engagement, rather than innovation with highly regarded specialists in Toronto, worrying about algorithmic and IT complexities.” Montreal and Edmonton attracting significant investments to those cities. A number of banks have also acquired companies in the AI space, in part as a talent grab to fuel CONSUMER PERSPECTIVES ON AI their own innovation activities,” the report stated. Consumer understanding and acceptance of AI is Much of the innovation is being applied to the marketing discipline, the focus of many AI startups. growing, due in part to the amount of popular culture According to “The Canadian AI Ecosystem: A 2018 references and media coverage of the burgeoning Profile,” conducted by Green Technology Asia, marketing technology. Greater appreciation of AI’s pervasiveness is the top industry judging by the number of AI firms has resulted from wide coverage of Facebook’s in Canada. The researcher found 38 AI startups with a platform exposures and the algorithms it uses to marketing focus, almost double the number of the next produce the newsfeed. Google’s much publicized largest industry, fintech. algorithm updates are also well covered. What this all means is consumers now better understand how AI underpins much of their interactions in daily life.

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 4 Global tech’s investment in Canada is rooted in the deep How Acceptable Are Select Uses of Artificial and culturally diverse talent pool within the country’s Intelligence (AI) and Algorithms According to Internet universities and startup community. And consumer Users in Canada? May 2018 % of respondents testing of AI attracts international players because of the AI video surveillance could determine and levy charges for 63% diversity within Canada’s urban centers full of digitally traffic violations savvy consumers. Could be introduced to your line of work and help you make 57% decisions that lead to better outcome or greater efficiency Could objectively examine the plans and policies of a political party 57% AI literacy has grown, but many consumers in Canada are and inform the public of the actual costs, benefits and likely less accepting of the technology than those in the US and unintended consequences offering voters more information UK. About a quarter of consumers in Canada don’t know Use AI instead of people to decide on shifting your investments into 52% new areas to maximize your returns how they feel about AI’s growth in the country, according AI could report sports news without human input 48% to a 2018 survey by OpenText. More than a third of Use AI instead of people to decide on air traffic, where and when 47% respondents (37%) didn’t know if they had interacted planes land at an airport with AI technology. AI video surveillance could determine and levy charges for serious 46% criminal offences like assault, robbery, etc. AI could report the political news without human input 44% Proof’s “CanTrust Index 2018” also shows internet users AI video surveillance refereeing sporting events* 42% in Canada are skeptical about AI. Fewer than four in 10 Use AI instead of people to decide health and dental insurance claims 36% respondents trust that AI will have a positive effect on Use AI instead of people to determine access to government support 33% the economy and customer experiences. This somewhat programs for such as employment insurance contradicts the Proof finding that 67% of respondents Use AI instead of people to decide which medical interventions are 32% best for you and your family during an illness or after an accident trust brands like Google, who are active in the AI arena. Use AI instead of people to decide implementation of government 28% policies such as who can immigrate to Canada Use AI instead of people to decide whether a person gets a job 22% Do Internet Users in Canada Trust Artificial AI could replace your job entirely 18% Intelligence to Have a Positive Effect on the Economy and Customer Experience? Feb 2018 Note: n=2,001 ages 18+; top 2 box "very acceptable" and "somewhat acceptable"; *e.g., game 7 of the Stanley Cup could see AI-driven % of respondents, by gender technology calling a penalty shot in the final minute of play Source: Ipsos and Communitech, "Tech for Good: A Canadian Perspective," Male May 23, 2018 42% 240539 www.eMarketer.com 41% The survey showed that people were more accepting of Female AI applications that augmented their decision making. 36% 34% They were neutral on applications that were lower touch, Total like reporting the news, and less accepting of applications 39% that had real-world consequences, as in AI medical 37% diagnosis and advice. Only 8% of respondents said they Positively contribute Improve customer were very knowledgeable about AI, while 12% said they to the economy experience were not at all knowledgeable. Note: ages 18+; top 3 responses on 7-point scale where 1=trust very little and 7=trust a lot Source: Proof Inc., "CanTrust Index 2018," April 10, 2018 Canada’s consumers aren’t as comfortable using 240612 www.eMarketer.com AI-powered chatbots as their counterparts in the US When asked about which AI use cases were acceptable, and Mexico. comScore and UPS surveyed digital buyers Canada internet users in response to an Ipsos and in each country about a range of chatbot utilities, and Communitech poll in May 2018 were more open to respondents in Canada were roundly less comfortable passive applications of AI—like video surveillance, which with getting product information, completing orders, topped the list, at 63%. When it came to replacing jobs returning products and receiving product updates. and other workplace issues, respondents were more resistant, at 18% and 22% respectively.

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 5 Digital Buyers in Canada, Mexico and the US Who Are “We’ve seen a significant shift with AI, particularly in Comfortable Using Chatbots for Select programmatic, from a performance standpoint,” said Shopping-Related Activities, 2017 Shawn Lowe, vice president of growth and partnerships % of respondents at Time and Space Media, a Halifax-based media agency. MexicoUS Canada “We would see eight to 10 campaigns generally across Getting product information 65% 42% 36% the span of a 12-month period, and we find that when an Completing my order 63% 39% 33% AI tech set is deployed, it continues to improve across New product updates 62% 39% 33% those different campaigns by up to 30%.” Ordering products 61% 38% 33% Returning products 58% 39% 30% Note: ages 18+; *includes Mexico For brand marketers, AI has also become relevant in a Source: UPS and comScore Inc., "UPS Pulse of the Online Shopper Study - few key areas that they can control. Personalization tops Global Study Executive Summary," April 11, 2018 237333 www.eMarketer.com the list, cited by 70.5% of business-to-business (B2B) marketers in North America in a Q1 2018 poll conducted The Canadian Artificial Intelligence Tracker, a 2017 survey by EverString and Heinz Marketing. Identifying trends of consumers conducted by Sklar Wilton & Associates, also rate highly, cited by 63.4% of respondents, as did found that many people have still-developing views on customization (58.1%), and processing data (55.1%). who the leaders are in AI. Almost half (43%) stated they However, the same survey showed that 13.7% of B2B did not know who a leading AI company was. Of those marketers in North America indicated that they did not who did have an opinion, Google topped the list, at “fully understand what AI is.” 29%, followed by Apple, Tesla, Microsoft, IBM, Amazon and Facebook. Areas in Which B2B Marketers in North America Are Most Interested in Using Artiﬁcial Intelligence to Support Their Marketing Strategy, Q1 2018 REAL-WORLD AI FOR MARKETERS % of respondents Personalization 70.5% The marketing industry has relied on and been Identifying trends 63.4% responsible for groundbreaking advances in AI. The Customization 58.1% much-publicized algorithms from Facebook, Google Processing data 55.1% and Amazon—to name an obvious few—are designed Pricing 20.7% to improve the relevancy of their services, make them 5.7% Other indispensable and monetize a targeted audience Source: EverString and Heinz Marketing, "The State of Artiﬁcial Intelligence for marketers. in B2B Marketing," June 28, 2018 239444 www.eMarketer.com The top digital advertising formats—search and display— For retailers, and ecommerce professionals in particular, rely on AI-based campaign optimization working beneath the utility of AI is clearer. Client-side ecommerce the surface. For example, the automated captions that professionals worldwide surveyed for the “Ecommerce appear on soundless Facebook ads are generated by AI Performance Report 2018,” by Econsultancy and tools. Programmatic ad channels are inherently machine Conversion, rated conversational commerce (e.g. chatbots learning-based, continually tuning bids to garner specified and personal assistants) as the top digital technology results for advertisers in display. they planned to experiment with by 2020. A full 59% of respondents said they would test it, and another 55% said they would utilize AI for personalization.

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 6 Digital Technologies that Agency* vs. Client-Side** “The overwhelming majority of developer and startup Ecommerce Professionals Worldwide Plan to interest is on Facebook Messenger in North America,” Experiment with by 2020 said Andy Mauro, CEO and co-founder of Automat.ai, a % of respondents Montreal-based conversational marketing developer. “We Agency* Client-side** get asked about WhatsApp now that it’s open. iMessage Conversational commerce (e.g., chatbots, 40% 59% personal assistants) is really interesting, but Apple’s careful with how they Artiﬁcial intelligence for personalization 39% 55% open it up. They’re making the same early mistake that Digital wallets/mobile payments 40% 44% Facebook made, which is they thought it was all about Social commerce 48% 40% customer service.” Voice technology 18% 18% Beacon technology 15% 15% Facebook reported that there were 8 billion monthly Different interfaces (e.g., smart watches, 14% 13% messages between users and businesses as of May internet of things) 2018, up fourfold from the year before. More than None of these 18% 16% Note: *includes ecommerce service providers (i.e., agencies, platform 300,000 bots are on the platform. implementers, software vendors); **includes client-side/brand marketers at ecommerce companies Source: Econsultancy and Conversion, "Ecommerce Performance Report Early testing of bots was primarily focused on a single 2018," Jan 18, 2018 use case: customer service, allowing a disaffected or 234767 www.eMarketer.com challenged customer to get a problem resolved quickly. “Retail globally is being transformed at every single While still viable, cost savings—reduced resourcing in level, and it’s at that perfect catching point to benefit a call center, for instance—rarely paid back adequately. from AI,” said Kerry Liu, CEO of Toronto-based Rubikloud Most recent chatbot development efforts are for engaging Technologies, which designs AI tools for retailers consumers during their purchase journey, answering worldwide. “The problems that retailers historically have product questions and facilitating a smoother, quicker around their business—like personalization—are really path to purchase. good problems for AI to come in and solve.” “The hype around chatbots is certainly way down,” Mauro said. “It’s very much like in the early days of email For more information on AI applications in marketing, read marketing, when people did not want to get emails eMarketer’s report, “Artificial Intelligence for Marketers from companies. You hear people saying that about 2018: Finding Value Beyond the Hype.” messaging now.”

According to Dan Code-McNeil, head of sales at Ada CHATBOTS Support, “AI-powered automation reduces agent wait times and allows companies to scale their offerings When Facebook opened up its Messenger platform to to match growth, while maintaining existing support developers in spring 2017, the notion of chatbots working budgets.” Understanding what metrics matter to for brands on messaging platforms had been tested, gauge the efficacy of a chatbot is maturing, and includes but not proven. Messaging platform Kik had tested bots the following according to Code-McNeil, whose since 2015, and there were many examples of bots Toronto-based firm has deployed chatbots for several working seamlessly in Asia on localized platforms, such customers in the market: as Line in Japan and WeChat in China. But the Facebook move brought the idea of conversational marketing via ■■ Engagement rate: ratio of conversations that were chatbots into the mainstream. Today, the vast majority engaged vs. abandoned of bot development and launches happen on Facebook

Messenger. (The company also announced in August ■■ Satisfaction rate: percentage of all reviews that 2018 an application programming interface (API) for were positive WhatsApp, which may lead to a more multi-platform approach to bots.) ■■ Recognition rate: percentage of questions the bot recognized and answered with anything other than “not understood”

■■ Handoff rate: percentage of conversations the bot hands off to an agent

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 7 As use cases expand, bot metrics have veered more “Other messaging platforms and voice interfaces on the toward demand generation aims like minutes per front end such as Google Home and Amazon Alexa will be engagement and average revenue per user. Automat.ai a part of the roadmap over time,” Tan said. published a Vichy case study in June 2018 showing some key metrics about the L’Oreal subsidiary’s bot Vicky, a virtual beauty consultant, which offers a diagnostic service CONCIERGE BANKING (the conversion trigger) based on skin conditions. Some results from the campaign on Facebook Messenger Financial services firms in Canada have taken the lead were compared with results of the diagnostic offered on in expanding chatbot capabilities into a fuller and more its website: curated “concierge” experience. Fintech AI firms have zeroed in on retail banking to augment how banks and ■■ Conversion: 87% of users took the chatbot’s diagnostic, financial institutions interact with customers. a rate four times higher than the website Concierge banking includes three primary use cases: ■■ Sales: 74% increase in items added to a user’s cart and retail banking, which includes products like lending, 35% greater revenue per user (defined as a user who mortgages, card services, fraud detection and branch took the diagnostic) compared with website results services; wealth management, including robo-advisors and investments; and insurance, using AI to help make “In the case of Vicky, in absolute numbers, we’ve seen sense of complicated products. up to 11 minutes of brand engagement,” Mauro said, whose firm developed the bot. “The average is in the “If you look at the consumer sector, you see the arrival 5- to 7-minute range.” After users receive their skincare of predictive, AI-based concierge services,” Flybits’ recommendations, the system informs back-end systems Rahnama said. “You have Siri on Apple, Google Assistant of consumer preferences to help with future offers and on Android, Cortana on Microsoft, so the question for follow-up content. a Fortune 500 company is how to build a predictive concierge service.” Like the beauty industry, airlines have been an early adopter of chatbots. In 2017, 78% of airline IT executives Consumers in Canada are heavy digital and mobile worldwide indicated they have implemented or plan to bankers, using financial services for checking account implement chatbots, according to Circle Research’s report balances, paying bills and purchasing additional services on airline IT trends. from lenders. According to GlobalWebIndex reporting on Q3 2017, 65.7% of internet users in Canada banked from WestJet became the first carrier in Canada to launch a a desktop/laptop, 40.9% from a mobile phone and 8.4% chatbot in August 2018. Dubbed Juliet, the bot enables from a tablet. flight shopping, itinerary information and flight booking. It’s also designed to handle customer support queries. Internet Users in North America Who Use Digital “The initial way we will gauge success hasn’t anything Banking, by Device, Q3 2017 to do with sales, marketing or revenue,” said Alfredo % of respondents Tan, chief digital and innovation officer, WestJet. “It’s the Canada speed at which we can process guest issues, deflecting 65.7% them from the call center.” 40.9% 8.4% WestJet estimates that “tens of thousands” of calls will US 49.3% be averted in the call center. The key metric that WestJet 49.6% tracked initially was percentage of queries handled by the 9.1% bot without the need to connect to the call center. The bot has been programmed to handle 700 query scenarios, Desktop/laptop Mobile Tablet according to Tan. Eventually, the company plans to expand Note: ages 16-64; in the past month Source: GlobalWebIndex, Feb 20, 2018 the AI tool beyond that of a messenger to its mobile app 235746 www.eMarketer.com and website.

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 8 Applying AI to highly sensitive personal information like Stange said simple transactions are well used, but there finances is a hard sell. People are ultra-conservative when is consumer education required to achieve that comfort it comes to money matters, and trust in AI is still low level with more complex transactions like mortgage among some segments of the population. To overcome adjustments and investments. After almost a year in these hurdles, the concierge apps from banks need to the hands of customers, the tool has provided lots of have super utility—using the best aspects of AI to make feedback on consumer attitudes and common issues them indispensable recommendation engines. discovered by the type of questions it fields.

“Our philosophy is that all marketing channels in the “We’re looking at a number of different use cases for future want to have an embedded recommendation AI, bots and robotics,” Stange said, whose company service,” Rahnama said. “Instead of the user searching for introduced a robot named “Pepper,” in branches in 2017 the information they want, the right information will find to field customer service questions. them when they need it. Companies can do this based on their own first-party data, and external sources like social The banking concierge tool was developed in partnership media data.” with Vancouver-based Finn.ai, which white labels virtual assistants for personal banking. The company’s solution The recommendation engine concept was popularized is being deployed by four Canadian financial institutions, by Amazon and Netflix. According to Rahnama, Amazon each with a unique twist. For BMO, the tool is being surfaces products and services based on about 12 data deployed to increase customer acquisition, for example. points, while the Netflix recommendation utility uses about five data points from viewership history. “The evolution of the space is going from very narrow use cases with chatbots to assistants that perform many In 2016, TD Bank launched TD for Me, a mobile app-based different use cases,” said Stephen Menon, a product experience that aims to curate personalized banking. leader at Finn.ai. “They now have the natural-language Today, almost 2 million of its customers are using the processing and deep-learning applications that allow them service, which is now powered by AI learning. to be much more intelligent and quicker to respond.”

To measure the efficacy of the app, TD relies on both These products are also moving beyond Facebook qualitative and quantitative metrics. The qualitative Messenger. The deployments from firms like Flybits ones include engagement rate and session times. The and Finn.ai are now operating within banking apps, quantitative metrics are more business related, and both native to Android and iOS and the mobile web. The include mortgage renewals, credit card upsells, insurance prospect of integrating voice capabilities in such tools is policy activation and fraud prevention. also imminent.

TD’s digital channels team—a mix of marketing and “There definitely is a shift to voice, but right now, we IT resources—developed on top of FlyBits’ software really are focused on text and natural language,” Menon development kit to bring the concierge capabilities to its said. “You can speak to your Siri, or any Android device, digital platforms, including its mobile app and website. and it will translate into natural language, and then In the end, TD decided to embed Flybits’ tech into its we can understand that translation. Because of the existing mobile app. complexity related to banking, it’s difficult to fulfill through a voice channel.” In Alberta, ATB Financial pioneered AI in western Canada in 2017 with the introduction of a banking concierge product, what it claimed to be the “world’s first fully featured virtual assistant.” SEARCH The world of search engine marketing (SEM) is being “We’re committed to offering alternative channels for upended by a variety of AI tools that include campaign consumers to do their banking in a frictionless way,” said optimization, predictive search and voice queries, enabled Curtis Stange, president and CEO of ATB. “The friction by natural-language processing. that exists in the movement of money is the trickiest to reduce.”

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 9 SEM Optimization Predictive Search (aka Recommendation Engines) Paid search efforts are traditionally straightforward, based Historically, search’s view to intent was clear, and ad on pay-per-click (PPC) pricing and defined budgets. Search products were simple and reliable. But predictive search ad spending in Canada will surpass US$2.9 billion in 2018, tools that recommend products or content based on according to our ad spending forecast. personalization are on many brands’ wish lists, especially those that sell online. SEM is done primarily through Google AdWords campaigns in Canada, as Google dominates search “Predictive search, which we define as the capability referral share in Canada with nearly 91% of the market, of a platform to anticipate what a consumer is most which is consistent with the worldwide average. likely looking for, is growing for all of our clients,” said Devon MacDonald, CEO of Mindshare Canada. “As “We were looking for a tangible value proposition media platforms learn more about consumer behavior associated with applying AI to our media-buying and intent through AI, there are incredible opportunities operations,” Time and Space Media’s Lowe said. to provide information and product recommendations “What we found was an AI overlay that sits on to consumers. Predictive search anticipates what the top of our AdWords account and optimizes PPC needs of the consumer are, based on their own and performance. We are seeing the doubling of campaign similar consumers’ previous behaviors, and in the process performance on impressions and clicks without any shortens the purchase cycle to a single view or click in incremental investment.” some instances.”

Time and Space is using Acquisio’s AI search tool, Almost two thirds (65%) of senior retail marketers reducing active workload by up to 40% for both small in North America and the UK surveyed by Monetate and large PPC client campaigns. The tedium of search in late 2017 said they were actively using a product campaign monitoring—measuring performance, bid recommendation engine. Only 11% of respondents said adjusting, budget capping—is offloaded to the automated they had no plans to implement one. AI system. Acquisio’s algorithms continually tweak search campaigns to fit within a certain budget and optimize for For retailers, recommendation engines “apply an AI lens specific days of the week that perform best. to customer engagement, taking hundreds of thousands of marketing offers and promotions at a retail level and “It’s self-learning bidding and budgeting for PPC,” determining the best combination of product offers and said Beth Thouin, vice president of marketing for messaging for an individual,” Rubikloud’s Liu said. Montreal-based Acquisio, which was acquired by Web. com in 2017. “It’s 24 hours a day, constantly adjusting so Predictive search doesn’t just apply to product that you’ll get the most clicks or conversions for a certain recommendations. It can also serve content from budget with never going over or under budget.” websites and apps for customers on the buying journey. Also known as “insight engines,” these AI-enhanced Google offers a homegrown cost per acquisition (CPA) content management tools suggest content for people algorithm that operates on budget parameters and PPC to view based on their profile and digital history with thresholds set up by advertisers. But third-party solutions the brand. that optimize PPC over time have become valued by media buyers like Time and Space Media. One of the Montreal-based Coveo was a pioneer in the indexing findings from the AI tool is that a lower PPC does not space, which organizes companies’ internal content and generate the best conversion outcomes. “If you pace it makes it discoverable with search. Advances in AI have and let the machine learn throughout the month, you’ll now made search predictive, and it has applications for end up with more conversions with the same budget both internal and customer-facing use cases. at the end of the month, plus consistent conversion,” Thouin said. “The industry has matured to where it is now, delivering predictive insights using machine learning and analytics,” said Diane Tetrault, senior director of product marketing at Coveo. “One of the primary use cases is for digital marketers and delivering personalized content from a website.”

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 10 Traditional websites are hard coded to present content Internet Users in Canada Who Use a Voice Assistant, based on what page a person surfs to on the site. With May 2017 & Feb 2018 predictive capabilities, the website can recommend % of respondents content based on user history or patterns in how they navigate a particular site. Coveo worked with Fasken, 25% a legal services firm, to make its website capable of 21% predictive search and recommended content. Fasken reported that time on its site increased 22%, a 32% increase in search utilization, a 46% increase clickthrough rate on suggestions made by machine learning and a 12% increase in registrations for webinars and email marketing. Voice Search May 2017 Feb 2018

The popularity of smart speakers like Google Home and Note: ages 12+; in the past week; e.g., Siri, Google Assistant, Cortana, Amazon Echo has led many marketers to consider a Alexa; via any device such as smartphone, smart speaker or connected TV Source: Solutions Research Group (SRG), "Digital Life Canada" as cited in voice strategy that includes search and conversational press release, April 16, 2018 marketing. Search terms are different when spoken 238233 www.eMarketer.com instead of typed, and brands need to understand the Specifically, smart speaker ownership was pegged at 8% semantic nuances. of consumers, based on a representative survey sample by Edison and Triton Digital, also from February 2018. For more information on how brands are using voice Google Home was the most popular brand of smart technology to increase engagement in, read eMarketer’s report, “Marketing Beyond the Screen: Using Voice speaker in Canada, owned by 55% of respondents in a Technology to Boost Brand Engagement.” May 2018 poll conducted by Media Technology Monitor. Fewer than a quarter (22%) of respondents owned an “Voice SEO will definitely be an important category in Amazon Echo, while 4% owned a speaker from Sonos making sure that you can have semantic data attached and 3% owned Apple’s HomePod, even though it hadn’t to your products and content so that you’re ranked officially launched in Canada at the time the survey was highly,” Automat.ai’s Mauro said. “Brands are increasingly conducted (It launched in Canada on June 18). thinking about it as conversational AI, as a layer to their marketing staff.” Smart Speaker Brands Owned by Smart Speaker Owners in Canada, May 2018 A quarter of internet users in Canada surveyed by % of respondents Solutions Research Group in February 2018 said they Google Home 55% used a voice assistant, which was defined as any device like a smartphone, smart speaker or connected TV with Amazon Echo 22% voice recognition. That was up from 21% in May 2017. 4% Sonos 3% Apple Homepod*

Note: ages 18+; *not yet launched in Canada at the time of the survey Source: Media Technology Monitor and CBC/Radio Canada, "Top 5 Sneak Peek," June 21, 2018 239128 www.eMarketer.com

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 11 While these penetration rates are still relatively low, ■■ Recommendation engines have become growth is expected and will have an impact on how instrumental in driving consumer actions, especially people search for products. One-third of worldwide ad in online retail. Amazon’s heavy investment in professionals responding to a Marin Software poll for personalization based on purchase history and other “The State of Digital Advertising 2018” report cited voice inputs has shown that the buyer journey can be search as a challenge that will impact their business accelerated with predictive capabilities. For digital this year. marketers, recommendation capabilities in delivering content has also developed to make website visits stickier. KEY TAKEAWAYS

■■ Brands in Canada are uniquely positioned to partner EMARKETER INTERVIEWS with AI innovators. As a key research country for machine learning, several startups are based in the How Financial Brands Create Virtual Assistants that market and are willing partners focused on making Respect Users’ Privacy digital marketing more effective. In cities like Montreal, Hossein Rahnama Toronto and Edmonton, the CA$125 million federal CEO, Founder Pan-Canadian Artificial Intelligence Strategy is funding Flybits collaboration between academia and the private sector. Interview conducted August 31, 2018

■■ Marketing is among the primary disciplines for WestJet’s Chatbot Is an Ongoing Project—Not a machine-learning research and development. Better One-Time Execution campaign management in search and programmatic display and customer engagement are just a couple Alfredo Tan of areas where brands can up their marketing game Chief Digital and Innovation Officer with AI. WestJet Interview conducted September 5, 2018

■■ Consumers in Canada are now more aware of their daily interactions with AI. As a result, they are Dan Code-McNeil becoming more open to increased reliance on the Head, Design technology on their phones (with virtual assistants like Ada Support Siri) and in their homes (via smart speakers like Amazon Interview conducted August 31, 2018 Echo and Google Home). For marketers, this growing consumer acceptance of AI means an opportunity Kerry Liu to design meaningful touchpoints that are mostly CEO automated and increasingly predictive. Rubikloud Technologies Interview conducted September 1, 2018 ■■ AI chatbots have become a popular testbed for automated routine consumer engagement. The Shawn Lowe vast majority of chatbots in North America have been Vice President, Growth and Partnerships deployed on Facebook Messenger, which was opened Time and Space Media up to developers in early 2017. Since then, brands Interview conducted August 22, 2018 have gone beyond using chatbots to handle customer service to more revenue-driving demand generation Devon MacDonald aims. In financial services, banks are developing CEO fuller concierge-type services via mobile apps and Mindshare Canada websites that simplify transacting and recommend Interview conducted September 13, 2018 complementary services. Andy Mauro CEO, Co-Founder Automat.ai Interview conducted August 21, 2018

ARTIFICIAL INTELLIGENCE INCUBATORS: WHY CANADA’S HUBS FOR AI RESEARCH ARE ATTRACTING GLOBAL TECH ©2018 EMARKETER INC. ALL RIGHTS RESERVED 12 Stephen Menon PwC Product Leader Proof Inc. Finn.ai Sklar Wilton & Associates Interview conducted August 30, 2018 Solutions Research Group Curtis Stange Triton Digital President, CEO UPS ATB Financial Interview conducted September 11, 2018 EDITORIAL AND Diane Tetrault Senior Director, Product Marketing PRODUCTION CONTRIBUTORS Coveo Interview conducted September 11, 2018 Joanne DiCamillo Senior Production Artist Katie Hamblin Chart Editorial Manager Beth Thouin Dana Hill Director of Production Vice President, Marketing Erika Huber Copy Editor Acquisio Ann Marie Kerwin Executive Editor, Content Strategy Interview conducted August 31, 2018 Stephanie Meyer Senior Production Artist Heather Price Managing Editor, Content Magenta Ranero Senior Chart Editor READ NEXT Amanda Silvestri Senior Copy Editor

Artificial Intelligence for Marketers 2018

SOURCES

CB Insights Circle Research Communitech comScore Conversion Econsultancy Edison Research EverString Heinz Marketing Ipsos Green Technology Asia GlobalWebIndex KPMG Marin Software MaRS McKinsey & Company Monetate OpenText

Coverage of a Digital World

eMarketer data and insights address how consumers spend time and money, and what marketers are doing to reach them in today’s digital world. Get a deeper look at eMarketer coverage, including our reports, benchmarks and forecasts, and charts.

Confidence in the Numbers

Our unique approach of analyzing data from multiple research sources provides our customers with the most definitive answers available about the marketplace. Learn why.

Customer Stories

The world’s top companies across every industry look to eMarketer first for information on digital marketing, media and commerce. Read more about how our clients use eMarketer to make smarter decisions.

Your account team is here to help: Email [email protected] to submit a request for research support, or contact [email protected] or 866-345-3864 to discuss any details related to your account. To learn more about eMarketer advertising and sponsorship opportunities, contact [email protected].